Trust Management for Digital Twins in Cloud-Based Environments

Digital Twins (DTs) are increasingly vital across sectors like healthcare and manufacturing, yet ensuring their runtime trustworthiness remains a challenge. This paper proposes a novel trust management model to dynamically evaluate a DT's reliability based on the impact of its insights on the physical entity's state. The model employs a voting mechanism between three analyzers namely Finite State Machine, Statistical, and Machine Learning to classify each insight as acceptable or unacceptable. Acceptable insights increase the DT's trust level and are executed, while unacceptable ones are blocked, mitigating security and performance risks. Demonstrated on an elevator system, this approach provides a foundational method for behavioral trust assurance in digital twin environments.